Crowd simulation is a complex research area related to a number of traditional computer science domains and also to psychology. This book describes algorithms and techniques for crowd simulation and is probably the best single-source document on multi-character simulation that I have found. It includes contributions from major researchers in this field.
The book has nine chapters. Chapter 1 is an introduction. Chapter 2 covers the state of the art of crowd simulation. The book is up to date as of April 2008, and this chapter could serve as a very good introduction to the area. Chapter 3 discusses issues related to the modeling of populations, from a perspective that focuses on creating diversity using multiple models. Chapter 4 presents animation of virtual humans; the section that discusses kinematic models is very interesting. Chapter 5 applies the information from the previous chapter to sociopsychological models of human movement in crowds. This is still an area where more research is needed, particularly when agents in a crowd have different motivations and goals, an aspect not fully covered in the chapter. Chapter 6 discusses the validation of the kinematic and avatar generation models by relating these to real crowds. The issue of tracking individuals is briefly discussed. Chapter 7 covers crowd rendering. This is an area that touches on the limits of computability; Thalmann and Musse present several adaptations to make crowd rendering a feasible computational task.
Chapter 8 describes the creation of virtual environments. This chapter provides details on the complexity of generating detailed realistic environments; several approaches are discussed. Some of the figures of the simulation tools used have captions in Portuguese, with no translation provided. This is the only flaw I have found in the book and it is quite irrelevant, as the comprehension of the figures is not compromised.
Finally, chapter 9 presents some case studies. The inclusion of a section on open issues and pointers to new research, as well as Thalmann and Musse’s vision of the future of crowd simulation, would have been very interesting.